Posting with data rather than just symptoms, since I’ve seen several similar threads here (1, 2, 3) without hard numbers attached.
Setup: 12x VTM31-SN (fw 1.1.5), HA 2026.9.2, ~6 Thread border routers (mostly Apple TV/HomePod + one Aqara Hub M100), one unified mesh (confirmed same partition ID and leader across every device).
Pulled HA’s Matter diagnostics export (ThreadNetworkDiagnostics cluster) for all 12 switches. Every single one shows real CCA (Clear Channel Assessment) failures — the radio finding the Thread channel busy when it tries to transmit:
-
Range: 14% to 115% of TX attempts hitting CCA failure (values over 100% reflect repeated retries)
-
Zero exceptions across all 12 units
-
No other Matter device in the house (Nest, Kasa, Aqara sensors, Govee) is even on Thread — only the Inovelli switches are, so they’re the only devices that could be generating this
This matches what @harryh’s thread found re: router-eligible devices auto-promoting (Thread trying to maintain 16-23 routers) and @francesc0’s multi-admin congestion theory — most of my switches are dual-fabric (Apple Home + HA), which doubles session overhead per device. Also matches @cpaholovach’s resolution path of 2.4GHz channel interference.
Questions for the group / Inovelli team:
-
Has anyone else pulled raw CCA/TxErrCcaCount numbers (attribute 36 on cluster 53) from their switches? Curious if this is universal at scale or specific to my RF environment.
-
Is there any router-eligibility control planned/available, or does Thread’s auto-promotion make that a non-starter by spec?
-
Any guidance on Thread channel selection for large (12+) White Series deployments specifically?